The GTS Group is pleased to announce that the CARTESIAN project has been awarded funding under the 2024 “Proyectos de Generación de Conocimiento y Actuaciones para la formación de personal investigador predoctoral” call from the Spanish State Research Agency (Agencia Estatal de Investigación). This recognition supports our commitment to advancing trustworthy, high-impact artificial intelligence and to training the next generation of researchers.

CARTESIAN focuses on developing advanced machine learning models that combine discrete latent representations with robust training strategies to address persistent weaknesses in generative AI and time-series modeling—most notably overconfidence and poor uncertainty calibration under domain shift. By integrating techniques such as error-correcting codes, adversarial training, and hybrid discrete–continuous architectures, the project aims to deliver models with stronger robustness, better generalization, and calibrated uncertainty estimates for real-world decision making.

Methodologically, the project introduces spectral-domain attention and new hierarchical designs to capture complex temporal dependencies, including non-stationarity and irregular sampling. These innovations will be applied to healthcare (behavioral monitoring and clinical outcome prediction), finance (models resilient to market complexity), sensor-array signal processing under imperfect conditions, and even the comparative analysis of historical canvases via latent probabilistic models. All models, datasets, and tools will be released openly to maximize reproducibility, collaboration, and impact across critical domains.